Today's Image & Video Generation: Fastest-Growing Projects — June 23, 2026
This week, the Image & Video Generation space continues to evolve rapidly with a variety of innovative projects emerging on GitHub. One notable trend involves more user-friendly interfaces and offline capabilities for AI image generation tools, catering to users who prefer privacy or have limited internet access. Additionally, there's an increasing focus on integrating multiple models and providing comprehensive workflows that span from concept creation to final output.
tianjiangqiji/nova-image-studio is a self-hosted AI image generation workbench that supports custom models and multi-mode operations including GIF generation and real-time task management. With its growth score of 40.75, it's gaining traction for offering a versatile platform with features like agent mode and adaptive UI layouts across different devices.
kadevin/ilab-gpt-conjure provides an AI image generation WebUI workbench specifically tailored to GPT-image-2, incorporating Codex Responses and OpenAI-compatible API support. This project has seen significant growth, accumulating 563 stars, likely due to its comprehensive feature set that includes a shared gallery, multi-task concurrency, and local queue management.
techjarves/Portable-Diffusion offers a portable AI image generator for Windows, Linux, and macOS, running GGUF and Safetensors models entirely offline with GPU/NPU acceleration. With 118 stars and a growth score of 23.64, this tool appeals to users who require an easy-to-use solution without the need for complex setup or system-wide dependencies.
wanshuiyin/ARIS-Movie-Director is designed for agentic long-horizon visual generation, converting fuzzy stories into cross-model-audited image-based movies. This project has a growth score of 22.68 and features robust multi-agent debate capabilities that enhance its appeal in the realm of multimodal content creation.
techjarves/Local-AI-Image-Generator is another fully self-contained, offline AI image generation studio for Windows, capable of running Stable Diffusion models locally with automatic GPU configuration. With 231 stars and a growth score of 17.53, it stands out due to its straightforward setup process that minimizes the need for manual intervention.
helloianneo/ian-xiaohei-scenes is a Codex Skill for generating Chinese real-object article illustrations and long-scroll story images with Xiaohei 2.0 capabilities. With 259 stars and a growth score of 8.87, it highlights the growing demand for localized content generation in specific cultural contexts.
nv-tlabs/dvlt is the official implementation of Déjà View: Looping Transformers for Multi-View 3D Reconstruction. Accumulating 339 stars with a growth score of 8.31, this project demonstrates significant interest from researchers and developers working on advanced visual reconstruction techniques.
calesthio/OpenMontage, despite having no recent commits or star ratings listed, is an ambitious open-source video production system that integrates over 400 agent skills across multiple pipelines and tools. Although growth data isn't available, the innovative nature of its agentic video production capabilities makes it a noteworthy entry in this category.
helloianneo/ian-xiaohei-illustrations, with an impressive 5,693 stars and a moderate growth score of 5.50, is another Codex Skill focused on generating hand-drawn illustrations for Chinese articles. Its popularity likely stems from its unique content generation capabilities tailored to the Chinese market.
Guo-chunyu/Chaning.G-s-Lrlab, with 100 stars and a growth score of 3.22, integrates large language models with neural network-level color grading algorithms for professional photography post-processing workflows. This project's growing interest is likely driven by its innovative approach to bridging AI capabilities directly into photo editing processes.
These projects collectively showcase the dynamic nature of image and video generation technology, offering a range of solutions from user-friendly interfaces to specialized tools designed for specific use cases or cultural contexts.
tianjiangqiji/nova-image-studio is a self-hosted AI image generation workbench that supports custom models and multi-mode operations including GIF generation and real-time task management. With its growth score of 40.75, it's gaining traction for offering a versatile platform with features like agent mode and adaptive UI layouts across different devices.
kadevin/ilab-gpt-conjure provides an AI image generation WebUI workbench specifically tailored to GPT-image-2, incorporating Codex Responses and OpenAI-compatible API support. This project has seen significant growth, accumulating 563 stars, likely due to its comprehensive feature set that includes a shared gallery, multi-task concurrency, and local queue management.
techjarves/Portable-Diffusion offers a portable AI image generator for Windows, Linux, and macOS, running GGUF and Safetensors models entirely offline with GPU/NPU acceleration. With 118 stars and a growth score of 23.64, this tool appeals to users who require an easy-to-use solution without the need for complex setup or system-wide dependencies.
wanshuiyin/ARIS-Movie-Director is designed for agentic long-horizon visual generation, converting fuzzy stories into cross-model-audited image-based movies. This project has a growth score of 22.68 and features robust multi-agent debate capabilities that enhance its appeal in the realm of multimodal content creation.
techjarves/Local-AI-Image-Generator is another fully self-contained, offline AI image generation studio for Windows, capable of running Stable Diffusion models locally with automatic GPU configuration. With 231 stars and a growth score of 17.53, it stands out due to its straightforward setup process that minimizes the need for manual intervention.
helloianneo/ian-xiaohei-scenes is a Codex Skill for generating Chinese real-object article illustrations and long-scroll story images with Xiaohei 2.0 capabilities. With 259 stars and a growth score of 8.87, it highlights the growing demand for localized content generation in specific cultural contexts.
nv-tlabs/dvlt is the official implementation of Déjà View: Looping Transformers for Multi-View 3D Reconstruction. Accumulating 339 stars with a growth score of 8.31, this project demonstrates significant interest from researchers and developers working on advanced visual reconstruction techniques.
calesthio/OpenMontage, despite having no recent commits or star ratings listed, is an ambitious open-source video production system that integrates over 400 agent skills across multiple pipelines and tools. Although growth data isn't available, the innovative nature of its agentic video production capabilities makes it a noteworthy entry in this category.
helloianneo/ian-xiaohei-illustrations, with an impressive 5,693 stars and a moderate growth score of 5.50, is another Codex Skill focused on generating hand-drawn illustrations for Chinese articles. Its popularity likely stems from its unique content generation capabilities tailored to the Chinese market.
Guo-chunyu/Chaning.G-s-Lrlab, with 100 stars and a growth score of 3.22, integrates large language models with neural network-level color grading algorithms for professional photography post-processing workflows. This project's growing interest is likely driven by its innovative approach to bridging AI capabilities directly into photo editing processes.
These projects collectively showcase the dynamic nature of image and video generation technology, offering a range of solutions from user-friendly interfaces to specialized tools designed for specific use cases or cultural contexts.